What can the RTX 5070 run?
| Text models: 246 | Visual models: 12 | Strongest first |
- Memory
- 12 GB
- Bandwidth
- 672 GB/s
- Maker
- NVIDIA
Open the calculator on this machineEvery model for 12 GB of VRAM
Text models that run on it
| Model | Best fit | Context | Speed | Strongest at |
|---|---|---|---|---|
| Qwen3.5 9B | 8-bit | 24K | 49 tok/s | Chat, Coding, Agents, Docs |
| Gemma 4 12B | 6-bit | 112K | 45 tok/s | Chat, Coding, Docs |
| Qwen3.5 4B | 8-bit | 176K | 86 tok/s | Agents, Docs |
| Ministral 3 14B | 4-bit | 16K | 55 tok/s | Coding, Agents, Docs |
| Gemma 4 E4B | 8-bit | 128K | 77 tok/s | Chat, Coding, Docs |
| gpt-oss 20B | 3-bit | 40K | 193 tok/s | Agents |
| Ministral 3 8B | 6-bit | 28K | 64 tok/s | Docs |
| Gemma 4 E2B | 8-bit | 128K | 137 tok/s | Chat |
| Nemotron 3 Nano 4B | 8-bit | 256K | 106 tok/s | Docs |
| Llama 3.1 8B | 8-bit | 20K | 56 tok/s | Chat, Coding, Agents, Docs |
| Qwen3 0.6B | 8-bit | 40K | 394 tok/s | Not rated |
| Qwen3 14B | 3-bit | 24K | 64 tok/s | Not rated |
| Qwen3 VL 8B Instruct | 6-bit | 24K | 65 tok/s | Text and images |
| Qwen3 8B | 8-bit | 16K | 55 tok/s | Not rated |
| Qwen2.5 7B Instruct | 8-bit | 32K | 59 tok/s | Not rated |
| Qwen2.5 VL 7B Instruct | 8-bit | 40K | 55 tok/s | Text and images |
| Unlimited OCR | 8-bit | 32K | 303 tok/s | Reading documents |
| Qwen2.5 1.5B Instruct | 8-bit | 32K | 231 tok/s | Not rated |
| Qwen3.5 2B | 8-bit | 256K | 170 tok/s | Text and images |
| Qwen3 VL 4B Instruct | 8-bit | 40K | 90 tok/s | Text and images |
| Qwen3 1.7B | 8-bit | 40K | 184 tok/s | Not rated |
| Qwen3 4B | 8-bit | 40K | 98 tok/s | Not rated |
| Qwen3 4B Instruct 2507 | 8-bit | 44K | 98 tok/s | Not rated |
| Qwen2.5 3B Instruct | 8-bit | 32K | 124 tok/s | Not rated |
| DeepSeek OCR | 8-bit | 8K | 303 tok/s | Reading documents |
| Qwen2.5 Coder 14B Instruct | 3-bit | 20K | 63 tok/s | Coding |
| Qwen2.5 VL 3B Instruct | 8-bit | 125K | 105 tok/s | Text and images |
| Qwen2.5 Coder 7B Instruct | 8-bit | 32K | 59 tok/s | Coding |
| Mistral 7B Instruct V0.2 | 8-bit | 28K | 59 tok/s | Not rated |
| Chandra OCR 2 | 8-bit | 160K | 77 tok/s | Reading documents |
| Qwen3 VL 2B Instruct | 8-bit | 72K | 177 tok/s | Text and images |
| Qwen2 VL 2B Instruct | 8-bit | 32K | 172 tok/s | Text and images |
| MiniCPM5 2B | 8-bit | 128K | 158 tok/s | Not rated |
| Gemma 3 4B IT | 8-bit | 128K | 109 tok/s | Text and images |
| Llama 3.2 3B Instruct | 8-bit | 64K | 124 tok/s | Not rated |
| LightOnOCR 2 1B | 8-bit | 16K | 321 tok/s | Text and images |
| Qwen2.5 14B Instruct | 3-bit | 20K | 63 tok/s | Not rated |
| Qwen3 1.7B Base | 8-bit | 32K | 212 tok/s | Not rated |
| HunyuanOCR | 8-bit | 96K | 308 tok/s | Text and images |
| Qwen3 0.6B Base | 8-bit | 32K | 457 tok/s | Not rated |
| Rax 4.5 | 8-bit | 256K | 170 tok/s | Text and images |
| Qwen2 VL 7B Instruct | 8-bit | 32K | 55 tok/s | Text and images |
| Ornith 1.5 9B | 8-bit | 20K | 49 tok/s | Not rated |
| Unlimited OCR 2 | 8-bit | 120K | 302 tok/s | Reading documents |
| Granite 4.1 3B | 8-bit | 88K | 113 tok/s | Not rated |
| DeepSeek R1 Distill Qwen 1.5B | 8-bit | 128K | 231 tok/s | Not rated |
| Mistral 7B V0.1 | 8-bit | 32K | 59 tok/s | Not rated |
| Qwen2.5 Coder 3B Instruct | 8-bit | 32K | 124 tok/s | Coding |
| Meta Llama 3 8B Instruct Abliterated V3 | 8-bit | 8K | 56 tok/s | Not rated |
| EXAONE 3.5 7.8B Instruct | 8-bit | 20K | 57 tok/s | Not rated |
| Saiga Llama3 8B | 8-bit | 8K | 56 tok/s | Not rated |
| Qwen2.5 Coder 1.5B Instruct | 8-bit | 32K | 231 tok/s | Coding |
| DeepSeek R1 0528 Qwen3 8B | 8-bit | 16K | 55 tok/s | Not rated |
| Gemma 3 12B IT | 5-bit | 28K | 55 tok/s | Text and images |
| Phi 3.5 Mini Instruct | 8-bit | 16K | 107 tok/s | Not rated |
| Qwen3 4B Thinking 2507 | 8-bit | 44K | 98 tok/s | Not rated |
| Phi 4 Mini Instruct | 8-bit | 52K | 104 tok/s | Not rated |
| Dots.ocr | 8-bit | 128K | 139 tok/s | Reading documents |
| DeepSeek R1 Distill Qwen 7B | 8-bit | 52K | 59 tok/s | Not rated |
| Phi 3 Mini 4K Instruct | 8-bit | 4K | 107 tok/s | Not rated |
Best fit is the highest quality, up to 8-bit, that leaves the use case’s context. Ranked by our use-case scores, then by Hugging Face downloads. Speeds are rough, plus or minus 25%. The top 60 of 246 are shown.
Image and video models
| Model | Best fit | Per output | Strongest at |
|---|---|---|---|
| Stable Diffusion XL 1.0 | 16-bit | 11 s | Images |
| Z-Image Turbo | 8-bit | 15 s | Images |
| FLUX.1 [dev] | 6-bit | 34 s | Images |
| Qwen-Image 2.1 | 8-bit | 4.0 min | Images |
| FLUX.3 Klein | 16-bit | 22 s | Images |
| FLUX.2 [klein] 9B | 8-bit | 7.7 s | Images |
| Krea 2 Turbo | 5-bit | 34 s | Images |
| Stable Diffusion 3.5 Large | 8-bit | 2.1 min | Images |
| MOSS-Image | 16-bit | 2.6 min | Images |
| Wan 2.2 TI2V-5B | 16-bit | 23 min | Video |
| MOSS Image Base | 16-bit | 2.6 min | Images |
| HunyuanVideo 1.5 | 6-bit | 1.9 h | Video |
Only just too big
- Qwen3.8 27B
- Muse Glimmer 30B
- Gemma 4 26B A4B
- Devstral Small 2 24B
- LFM2.5 24B A2B
- Mistral Small 3.2 24B
- MiniMax H3
- LTX-2.5
- Qwen3.6 27B
- Qwen3.5 27B
- Diffusiongemma 26B A4B IT
- Gemma 3 27B IT
- Qwen-Image
- GPT Neox 20B
- Wan2.2 I2V A14B
- CodeLlama 7B HF
- Wan2.1 I2V 14B
- Wan 2.1 T2V-14B
- Qwen1.5 7B
- Deepseek Coder 6.7B Base
- Qwen 7B Chat
- Qwen Image 2512
- Molmo2 O 7B
- Internlm2 Chat 20B
- Qwen Image Bench
- Internlm2 Chat 20B ExPO
- Medgemma 27B Text IT
- Qwen1.5 14B Chat
- CodeLlama 13B Instruct HF
- Wan 2.2 A14B
These need up to a quarter more memory than the RTX 5070 has. Spilling into system RAM or a lower precision can get them running, slowly.
Questions
What is the best LLM for the RTX 5070?
Of the open models we track, the strongest that run well on the RTX 5070 are Qwen3.5 9B, Gemma 4 12B, Qwen3.5 4B. 246 text models fit it in all.
Calculated from published specs, not benchmarked. All machines